Remember me
Chronic noncommunicable diseases (NCDs) represent a significant health challenge worldwide, imposing an increasing economic burden on health care services and resulting in premature deaths [,]. To reduce the burden of NCDs, reducing modifiable lifestyle risk factors has been highlighted as one of the most important actions worldwide []. Adopting a healthy lifestyle can potentially reduce the risk of developing common chronic diseases by up to 80% []. There is a pressing need to develop strategies that promote the widespread adoption of healthy lifestyle habits, aiming to achieve impactful results at a large scale.
Digital health interventions (DHIs) offer a viable way to support a healthy lifestyle and hold promise as a low-cost means of delivering interventions to large population groups. However, DHIs can be effective only if they reach the target population, especially those who need them most. Ensuring that DHIs reach their intended users requires awareness of the potential barriers to uptake, which has been broadly defined as the initial step of registering for, downloading, or otherwise beginning to use a DHI [,]. This is particularly important in the context of NCDs, which disproportionately affect populations with lower socioeconomic status and limited access to health care services [,]. DHIs have the potential to improve access to preventive care and support individuals in managing their health, but to truly make a difference, it is essential to remove barriers that could otherwise deepen existing health inequalities []. Currently, there is a lack of comprehensive understanding regarding the factors influencing DHI uptake. Identifying these factors will facilitate the customization of future interventions and recruitment strategies, thereby enhancing adoption rates and promoting digital equity.
Digital health promotion is an emerging field, and therefore, research on DHIs remains scarce []. Much of the existing literature has concentrated on health apps, which often lack a defined duration, evidence-based modules, and explicit behavior change techniques that are typically built into DHIs [,]. These studies have shown that females [,,], young or middle-aged individuals [,,], and those with higher education levels [,,] and higher income [,,] tend to be more frequent users of health apps. Additionally, conflicting findings exist regarding the relationship between overall health [,,], lifestyle habits [,,,], and health app usage. Recent research has also suggested that additional factors, such as perceived usefulness and privacy concerns, might play a significant role in influencing the adoption of health apps and digital health services [-]. Perceived usefulness has been associated with greater acceptance and intention to use digital health services across age groups [,]. In contrast, concerns regarding privacy and data security, particularly among older adults, can reduce usage intentions and act as barriers to adoption [-].
Studying health app users, although informative, does not offer comprehensive insights into potential barriers to DHI uptake. First, the lack of comparative analysis between adopters (individuals who start using DHIs) and nonadopters raises questions about whether observed characteristics reflect a higher uptake probability or the characteristics of the populations to which the apps were advertised. Second, the usage of health apps itself may influence individuals’ lifestyle habits and health [], introducing a potential bias and thereby complicating the interpretation of these findings in relation to DHI adoption. Lastly, usage is also influenced by other factors, such as user experience, which may be affected by the app quality [].
Here, our aim was to study factors predicting DHI uptake to identify potential groups that may be less likely to start using DHIs. We examined whether socioeconomic factors, self-reported health and lifestyle habits, or habitual use of electronic services (e-services) predicts DHI uptake. Understanding these factors will help identify population groups that are underrepresented in DHIs and find alternative approaches for them. These findings will also aid in integrating potential barriers into the formulation of recruitment strategies and the development of DHIs themselves.
This cross-sectional study was a substudy of the Healthy Finland population survey, the methodology of which is described in detail elsewhere [,]. In this survey, a questionnaire on health, well-being, and service use was sent to randomly selected individuals over the age of 20 years, representing the entire adult population of Finland. For our study, Finnish-speaking individuals aged 20-74 years who had completed the Healthy Finland questionnaire between September 2022 and the end of 2022 were considered eligible, excluding those randomly selected for the health examination component to prevent potential influence of the DHI on examination results. Among this eligible population (N=10,207), an SMS invitation was sent to all individuals with a known phone number (n=4978, 48.8%). Additionally, 2000 (19.6%) individuals from the remaining eligible population were sampled based on 5-year age groups to receive a letter invitation. Three participants asked their data to be removed, resulting in a final sample size of 6975 (99.9%) individuals. The sample size was chosen to detect small effects in logistic regression, with a statistical power of 0.8 and a type I error of 0.05, assuming that approximately 10%-15% of invitees would adopt the DHI. Based on this, a minimum total sample size of 2538 participants was required.
Participants were offered the opportunity to use a DHI app (BitHabit) for 3 months []. The invitation letter included a web address, and the SMS message contained a direct link to the project’s web page (hosted by the Finnish Institute for Health and Welfare), where participants could find information regarding the study, a link to the BitHabit app, and brief instructions on how to get started with the app.
Ethical ConsiderationsEthical approval for the study was obtained from the research ethics committee of the Finnish Institute for Health and Welfare (THL/5335/6.02.01/2022). The approval covered all the procedures, data, and analysis presented in this study. Participation in the Healthy Finland survey was voluntary, and participants were informed that completing and returning the questionnaire constituted consent to participate. In addition, digital informed consent was obtained following BitHabit registration. Participants were informed of their right to decline or withdraw participation at any time. All data were pseudonymized before analysis, and the study was conducted in accordance with the General Data Protection Regulation (GDPR) and applicable Finnish data protection legislation and the analysis conducted in a secure analysis environment. No compensation was provided to participants.
The Digital Health Intervention App (BitHabit)The BitHabit app was originally developed to support the formation of healthy lifestyle habits in adults at increased risk of type 2 diabetes [,]. The app includes a broad selection of small health-promoting actions (“habits”) in 14 different categories related to physical activity, diet, sleep, stress management, positive mood, smoking, and alcohol consumption. It is a web-based app that can be used on a mobile phone, tablet, or computer. To log into the app, participants had to provide a phone number and a user ID that was specified either in the letter or in the SMS message. Uptake of the app was defined as the process of registering for the BitHabit app with a phone number and user ID, agreeing to the terms of use, and, subsequently accepting the invitation to participate. No additional confirmation of refusal was collected from those who did not register.
Predictor VariablesWe examined predictors from the following categories: socioeconomics, health, lifestyle habits, and the use of e-services. All predictor variables except age and sex (assigned at birth), which were obtained from the Finnish National Population Register, were derived from the Healthy Finland questionnaire and were thus self-reported. The survey was conducted approximately 2-4 months prior to sending invitations to take part in the DHI substudy, and the BitHabit app was offered only after the survey responses had been received. Following the invitation, participants had approximately 4 weeks to begin using the app. All questions and the answer options are presented in .
SocioeconomicsSocioeconomic variables included age, sex, education (years of education, including primary and comprehensive school), household income (€ per year before taxes), and employment status. Household income was coded as a five-category ordinal variable: <€15,000 (<US $17,437; calculated at an exchange rate of €1=US $1.16 approximately), €15,001-35,000 (US $17,438-$40,686), €35,001-55,000 (US $40,687-$63,935), €55,001-75,000 (US $63,936-$87,184), and >€75,000 (>US $87,184). Age was categorized into four groups: 20-34, 35-49, 50-64, and 65-74 years.
HealthThe BMI was calculated based on self-reported height and weight. Other health-related predictor variables included functional capacity, ability to work, current health status, prevalence of (any) long-term illness, health care service use (number of visits during the past 12 months), perceived ability to learn, and perceived limitations due to health problems. Functional capacity was calculated as a summary score for five questions on the abilities to perform the following activities: run for about 100 m, walk for about 500 m without stopping to rest, see ordinary newspaper print, hear what is said in a conversation between several people, and walk up one flight of stairs without stopping to rest. Each subquestion had four choices (yes, with no problem [3 points]; yes, with some difficulty [2 points]; yes, but with great difficulty [1 point]; no, I cannot [0 points]). A higher summary score indicated better functional capacity.
Lifestyle HabitsLifestyle-related information included physical activity, sleep, smoking, and diet quality. Participants were divided into two groups based on whether they achieved the Finnish physical activity recommendations []. Sleep adequacy categories were created based on the question, “Do you get enough sleep?” The diet quality score was calculated following the method presented by Lindström et al [], and the details of the scoring are in the Diet Quality Scoring section in .
Use of e-ServicesThe following predictors relating to the use of e-services were considered: independence of e-service use, competence to use e-services, accessibility of e-services, concerns about data security, poor internet connections, and perceived benefits of e-services. The “perceived benefits of e-services” score was a summary measure calculated based on participants’ level of agreement with six claims: e-services (1) help me assess the need for services, (2) support me in finding and choosing the most suitable services, (3) make it easier for me to use the services regardless of where I am and when, (4) make it easier for me to collaborate with professionals, (5) help me take an active role in looking after my own health and welfare, and (6) help me take care of the health, welfare, and functional capacity of family or friends. Each claim had five categories (completely agree, somewhat agree, neither agree nor disagree, somewhat disagree, strongly disagree). A higher score indicated that the participant perceived e-services as more beneficial.
Statistical AnalysisLogistic regression models were used to assess the association between the predictor variables and the uptake of the BitHabit DHI. The results were presented as adjusted odds ratios (aORs) with 95% CIs. Models examining variables related to health, lifestyle, and use of e-services were adjusted by contact method (SMS or letter), age, sex, education, and income. Regarding socioeconomic factors, models for age, sex, and contact method were adjusted by each other; models for income, BMI, and employment status by education, contact method, age, and sex; and the model for education by contact method, age, and sex. These covariates were selected a priori as background variables likely to be confounders of the predictor-outcome relationship based on the previous literature. The only exception was contact method (SMS vs letter), which was included to adjust for potential confounding introduced by the recruitment strategy: letter invitations were sampled by age to ensure representation across age groups, potentially correlating with both participant characteristics and DHI uptake. Participants with missing data (missing answers) were excluded from the analyses, and a complete-case analysis was performed. The amount of missing data for each predictor variable is presented in the Results section.
The overall significance of the associations between the categorical predictors and the outcome was assessed with likelihood ratio tests (LRTs), for which a false discovery rate (FDR)–corrected P value of <.05 was determined to indicate a significant association. R software version 4.3.1 (R Foundation for Statistical Computing []) was used to perform all statistical analyses. The reporting of this study followed the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines for reporting observational studies [].
The final sample of 6975 participants, 1287 (18.5%) started using the BitHabit app (“adopters”). The uptake proportion was 16.2% (806/4975) among those who were invited via SMS and 24.1% (481/2000) among those with a letter invitation. Those invited via SMS were less likely to start using the app than those invited via mail (aOR 0.65, 95% CI 0.55-0.75). Background information for the entire study sample and separately for app adopters and nonadopters is presented in -.
Table 1. Socioeconomic characteristics of the study sample.aCharacteristicsTotal sample (N=6975)Adopters (n=1287)Nonadopters (n=5688)Age (years), n (%)aData are reported as the mean (SD) for continuous variables and as n (%) for categorical variables.
bAn exchange rate of €1=US $1.16 approximately was applied.
cThe number of participants was as follows: total, N=4067; adopters, n=761; and nonadopters, n=3306.
Table 2. Health-related characteristics of the study sample.aCharacteristicsTotal sample (N=6975)Adopters (n=1287)Nonadopters (n=5688)BMI (kg/m2)aData are reported as the mean (SD) for continuous variables and as n (%) for categorical variables.
Table 3. Lifestyle-related characteristics of the study sample.aCharacteristicsTotal sample (N=6975)Adopters (n=1287)Nonadopters (n=5688)Diet qualityaData are reported as the mean (SD) for continuous variables and as n (%) for categorical variables.
Table 4. E-servicea use–related characteristics of the study sample.bCharacteristicsTotal sample (N=6975)Adopters (n=1287)Nonadopters (n=5688)Independence of e-service useae-service: electronic service.
bData are reported as the mean (SD) for continuous variables and as n (%) for categorical variables.
Predictor Variables Associated With DHI UptakeFactors associated with the uptake of the BitHabit app were studied in four domains: socioeconomics, health, lifestyle habits, and use of e-services. The aORs from all the models, along with associated P values and 95% CIs, are listed here, while details of the adjustments for each variable are listed in Methods section.
SocioeconomicsAge, sex, education, and household income demonstrated a significant association with uptake of the BitHabit DHI ( and ). Odds of adopting the DHI were higher for individuals aged 35-49 years compared to the reference group of 20-34 years (aOR 1.47, 95% CI 1.21-1.79). The two older age groups (50-64 and 65-74 years) showed no significant difference from the reference group. Higher educational attainment was associated with greater odds of DHI uptake (aOR 1.10, 95% CI 1.08-1.12). Similarly, higher income levels were associated with increased uptake, though the two highest categories had comparable estimates (€35,001-55,000: aOR 1.76, 95% CI 1.37-2.27; €55,001-75,000: aOR 1.97, 95% CI 1.51-2.59; >€75,000: aOR 1.96, 95% CI 1.50-2.59). In addition, women were more likely than men to adopt the DHI (aOR 1.69, 95% CI 1.49-1.93), whereas employment status was not significantly associated with DHI uptake.
Figure 1. Forest plot showing aORs and 95% CIs for socioeconomic factors. Each aOR is adjusted for relevant covariates, as detailed in the Statistical Analysis section. aOR: adjusted odds ratio; ref: reference. Table 5. Logistic regression and LRTa results for socioeconomic predictor variables.Predictor variablesLRTLogistic regressionaLRT: likelihood ratio test.
bFDR: false discovery rate.
caOR: adjusted odds ratio.
dNot applicable.
eAn exchange rate of €1=US $1.16 approximately was applied.
HealthAmong the health-related predictors, functional capacity and current health status showed significant associations with DHI uptake ( and ). The odds of adopting the DHI were higher for individuals with better functional status (aOR 1.06, 95% CI 1.02-1.11) and better current health (good or fairly good compared to poor: aOR 1.46, 95% CI 1.15-1.89). In contrast, none of the other health indicators, including the BMI, ability to work, ability to learn, the presence of long-term illnesses, the number of health care visits during the past 12 months, and limitations due to health problems, were significantly associated with uptake of the DHI.
Figure 2. Forest plot showing aORs and 95% CIs for health-related predictors. Each aOR is adjusted for relevant covariates, as detailed in the Statistical Analysis section. aOR: adjusted odds ratio; ref: reference. Table 6. Logistic regression and LRTa results for health-related predictor variables.Predictor variablesLRTLogistic regressionaLRT: likelihood ratio test.
bFDR: false discovery rate.
caOR: adjusted odds ratio.
dNot applicable.
Lifestyle HabitsDiet quality, smoking, and sleeping were associated with the uptake of the DHI ( and ). Participants with better diet quality (aOR 1.07, 95% CI 1.04-1.10) had higher odds of DHI uptake. Similarly, smoking behavior showed a clear gradient: Compared to daily smokers, those who smoked occasionally (aOR 1.59, 95% CI 1.08-2.36), did not smoke at the time of the survey (aOR 2.26, 95% CI 1.69-3.06), or had never smoked (aOR 2.29, 95% CI 1.72-3.12) had higher odds of adopting the DHI. Sleeping enough was also a significant predictor; however, no meaningful difference was estimated between those sleeping mostly enough and the reference category of those sleeping mostly not enough. Instead, participants who were unsure of whether they slept enough had significantly lower odds of DHI uptake compared to the reference category (aOR 0.58, 95% CI 0.37-0.86). In contrast, achieving physical activity recommendations was not significantly associated with uptake of the DHI.
Figure 3. Forest plot showing aORs and 95% CIs for lifestyle habits. Each aOR is adjusted for relevant covariates, as detailed in the Statistical Analysis section. aOR: adjusted odds ratio; ref: reference. Table 7. Logistic regression and LRTa results for lifestyle-related predictor variables.Predictor variablesLRTLogistic regression
Comments (0)